Time-Dependent Behavior of Continuous Composite Integral

Abutment Bridges
M. Arockiasamy, P.E., M. ASCE,1 and M. Sivakumar2

Abstract: Integral abutment bridges provide bridge engineers with an economical and attractive design alternative to traditional bridges
with thermal expansion joints. Very limited design and construction guidelines are available and no unified design procedures exist, hence
there is a lack of enthusiasm to adopt integral abutment bridges for long spans. Time-dependent effects of creep, shrinkage, and
temperature on composite superstructure, soil pressure from abutment backfill, and soil–substructure–superstructure interaction are the
main parameters governing the design. Analytical models and numerical procedures are developed for predicting instantaneous linear and
nonlinear time-dependent long-term behavior of continuous composite integral abutment bridges. The redistributions of moments due to
temperature gradient, creep, shrinkage, and restraints provided by abutment foundation and backfill have been considered in the analysis.
The substructure system is modeled as discrete springs for translational and rotational degrees of freedom. A numerical design example
is illustrated showing the deformations and stresses due to time-dependent effects on a continuous composite deck due to typical sustained
loads and compared with those of the conventional bridge system on seat-type abutments. The axial forces and moments on the piles
supporting the abutment were used to analyze the behavior of the laterally loaded piles. Recommendations are presented for the analysis
and design of integral abutment bridges.
DOI: 10.1061/共ASCE兲1084-0680共2005兲10:3共161兲
CE Database subject headings: Bridge abutments; Bridges, continuous; Bridges, composite; Creep; Shrinkage; Temperature effects;
Pile lateral loads; Time dependence.

Introduction column or shafts 共Lam and Martin 1986兲. The integral bridges
differ from regular rigid frame bridges in the distribution of
Integral construction is used to avoid problems associated with stresses due to temperature change, prestressing, creep, shrinkage,
bridge deck joints and reduce the construction and maintenance and restraints provided by abutment foundation and backfill. Con-
costs. A continuous jointless deck integrally connected to the tinuity between the abutment and the superstructure results in
abutment is one of the major types of construction of integral secondary stresses. Time-dependent effects of creep, shrinkage,
abutment bridges. The end diaphragm or the abutment is cast and temperature on composite superstructure, soil pressure from
monolithically with the superstructure and may be directly sup- abutment backfill, and soil–substructure-superstructure interaction
ported on piles. The structural components of a typical integral are the main parameters governing the design. The behavior of
bridge consist of superstructure, abutment, abutment footing con- integral abutment bridges is also influenced by the
sisting of a single row of piles, abutment backfill, and wing walls. superstructure–substructure stiffness, type of foundation, connec-
A typical integral abutment bridge is shown in Fig. 1. Integral tion details between the foundation and abutment, soil properties
abutment bridges accommodate superstructure movements with- and soil settlement, and the connection between the approach slab
out conventional expansion joints. One of the advantages of inte- and the deck system. Since only very limited design and construc-
gral construction is better seismic resistance due to the added tion specifications are available in the American Association of
redundancy. The longitudinal earthquake-induced inertial forces State Highway and Transportation Officials Specifications
are transmitted directly from the superstructure to the soil behind 共AASHTO 1998兲, wide variations are found among various states
the abutment without having to pass through bearing devices. in idealizing integral abutment bridges for analysis and design
Such abutments have performed well during earthquakes and 共Kunin and Alampalli 2000兲.
avoided problems such as backwall and bearing damage with Although the majority of the bridges with integral abutments
seat-type abutments and also reduced the lateral load taken by perform adequately, and the stress levels generated by secondary
effects are well understood from the field observations and per-
1
Professor and Director, Center for Infrastructure and Constructed formance 共Shen and Lopez 1997兲, many of them operate at high
Facilities, Florida Atlantic Univ., 777 Glades Rd., Boca Raton, FL 33431. stress levels 共Burke 1990兲. In spite of the fact that integral bridges
2
Design Engineer, PTE Strand Co. Inc., Hialeah, FL 33014. have acceptable performance under seismic loads, or under long-
Note. Discussion open until January 1, 2006. Separate discussions term effects due to creep, shrinkage, and temperature, there is no
must be submitted for individual papers. To extend the closing date by comprehensive model available to analyze the behavior. Design
one month, a written request must be filed with the ASCE Managing
and construction practices are based largely on past local experi-
Editor. The manuscript for this paper was submitted for review and pos-
sible publication on December 11, 2003; approved on July 16, 2004. This ence and intuitive practice rather than systematic investigation
paper is part of the Practice Periodical on Structural Design and Con- and thus are empirical in nature. Although the effects of creep and
struction, Vol. 10, No. 3, August 1, 2005. ©ASCE, ISSN 1084-0680/ shrinkage in a composite girder were investigated on continuous
2005/3-161–170/$25.00. concrete bridges with simple supports 共Gilbert and Bradford

PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005 / 161
 Fig. 1. Typical integral abutment bridge

1995; Kwak and Seo 2000; Destrebecq and Jurkiewiez 2001兲, integral abutment bridge involve their ability to accommodate
only limited publications are available on detailed analytical pro- potentially large superstructure movements and the sharing of lat-
cedures to analyze the continuous composite integral abutment eral and longitudinal forces among the substructure units. As part
bridge under sustained service loads. An analytical procedure in- of the overall structural system, integral abutment bridge piers
corporating the time-dependent behavior of the superstructure will typically be required to carry a portion of the externally
will aid in determining the forces transmitted to the substructure applied longitudinal and transverse loads on the bridge.
system of the integral abutment bridge. In the abutment and pile design, the girders transfer to the
The substructure system of an integral abutment bridge typi- foundation all moments, vertical, and horizontal forces that are
cally consists of integral abutment, wing wall, backfill behind the produced by the superimposed dead load, live load with impact
abutment, and flexible piles supporting abutments. Since the su- factor, earth pressure, temperature, shrinkage, creep, and seismic
perstructure of the integral abutment bridges does not have con- loads. The piles can be provided with predrilled holes, if the
ventional expansion joints, the substructure system offers a re- stresses developed in the laterally loaded piles are greater than the
straint to the superstructure movement. This restraint should be allowable stresses. Where piles are driven in dense and stiff soils,
considered in modeling the superstructure. A parametric study preaugured holes filled with loose sand can be provided to reduce
using two-dimensional frame analyses was reported resistance to lateral movement.
共Thippeswamy and GangaRao 1996兲 on the effects of the super- Approach slabs should be provided in order to prevent vehicu-
structure and substructure stiffness in the distribution of moments lar traffic from consolidating the backfill adjacent to abutments, to
and axial forces in the pile. The stiffness of the substructure sys- eliminate live load surcharging of backfill, and minimize the ad-
tem includes the contribution from abutment walls, pile founda- verse effect of consolidating backfill and approach embankment
tion, and soil 共Wilson 1988兲. This has been achieved by modeling on movement of vehicular traffic. Cycle control joints, which fa-
the system as discrete springs for translational and rotational de- cilitate longitudinal cycling of bridges and approach slabs, should
grees of freedom. be provided between approach slabs and approach pavement. The
backfill material behind the abutment wall should accommodate
Design and Analysis Requirements the lateral movement of the bridge without producing large
stresses to the bridge system.
Integral abutment bridges are designed to resist all the vertical
and lateral loads. The load effects at various stages of construc-
tion should be considered in the design. The construction stage at
which the structure is simply supported, and the stage of making Theoretical Model
it integral with abutments and backfill, are of primary importance.
The maximum length of integral abutment bridges is deter- The analysis of a continuous composite integral abutment bridge
mined by the soil properties, seasonal temperature variations, re- under sustained service loads should consider the inherent prop-
sistance of abutment foundations to longitudinal movements, i.e., erties of the constituent materials, namely, structural steel and
the ability of the piles supporting the abutments to withstand the concrete. Realistic evaluation of internal forces, support reactions,
bending moments and shear forces resulting from the movement stress and strain, together with cracking of concrete and deflec-
of the abutments, and the type of superstructure being considered. tions, is required to check the adequacy for the serviceability limit
Seasonal temperature variations causing expansion/contraction state. In order to achieve this, it is necessary to consider the
mobilize the horizontal soil pressure. Owing to the nonuniform difference in rheological behavior between the constituent mate-
distribution of loads and difficulties in establishing the movement rials, cracking of concrete in negative bending moment region,
and its direction, structures with skews greater than 35° or where and the construction process. Unlike steel, the evolution of me-
an angle subtended by a 30 m arc-length of the structure is greater chanical characteristics of concrete due to time-dependent shrink-
than 5° are not considered suitable for integral abutment designs. age and creep strains causes the structure to undergo deflections
Piers for integral bridges have similar design requirements and and redistribution of stresses. Cracking of concrete under tension
share common design procedures with those of a more traditional reduces the stiffness of the structural element and thereby initiates
bridge. The primary distinguishing features of the piers for an a modified stress distribution among the structural elements. In

162 / PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005
 ␴c共t0兲
␧c = 关1 + ␾共t,t0兲兴 共1兲
Ec共t0兲
where ␾共t , t0兲⫽ratio of creep strain to instantaneous strain and a
function of the age at loading, t0, and the age t for which the
strain is calculated; ␴c共t0兲⫽sustained concrete stress; and
Fig. 2. Modeling of superstructure Ec共t0兲⫽modulus of elasticity of concrete at age t0. This linear
relationship, which is true within the range of stresses under sus-
tained loads, allows superposition of the strain due to changes in
stresses and shrinkage. Thus, the total strain in concrete due to
applied stress and shrinkage is given by
the case of continuous bridges, it also affects the redistribution of

冕
internal forces. When the superstructure, consisting of concrete/ ⌬␴c共t兲
steel girders and concrete deck slab, is subjected to sustained 1 + ␾共t,t0兲 1 + ␾共t,␶兲
␧c共t兲 = ␴c共t0兲 + d␴c共␶兲 + ␧cs共t,t0兲
loads, it undergoes nonlinear time-dependent deformations in the Ec共t0兲 0 Ec共␶兲
concrete due to the nonlinear temperature variation across the 共2兲
cross section, creep, and shrinkage, in addition to cracking of the
concrete. where ␧cs共t , t0兲⫽free shrinkage strain occurring between ages t0
The integral abutment bridges do not have expansion joints and t; and ⌬␴c共t兲⫽gradual increment in concrete stress intro-
between the abutments and the bridge superstructure; therefore, duced during the period t0 to t. A reduced creep coefficient can be
the bridge girders are rigidly connected to the abutments. The used to calculate the creep strain, if the stress is applied gradually.
substructure generally consists of stub type abutment supported With this simplification, the integral equation can be eliminated
on a single row of piles. Bridge movements caused by the thermal and Eq. 共2兲 can be modified as
expansion or contraction, creep, and shrinkage of concrete induce
bending stresses in the bridge system. The induced bending 1 + ␾共t,t0兲 1 + ␹␾共t,t0兲
␧c共t兲 = ␴c共t0兲 + ⌬␴c共t兲 + ␧cs共t,t0兲 共3兲
stresses must be resisted by the bridge superstructure and sub- Ec共t0兲 Ec共t0兲
structure. Even though the superstructure is rigidly connected to
where ␹⫽aging coefficient.
the abutment in order to transfer the induced moment due to
For any applied moment M i and axial force Ni, the instanta-
temperature, creep, and shrinkage to the abutment pilings, many
neous strain distribution ␧oi and curvature ␬i are obtained as
state Department of Transportations treat the integral abutment
bridges as simply supported due to simplicity in the analysis. The
integral abutments may be considered as rigid owing to their
physical configuration and restraint provided by the backfill.
再冎
␧oi
␬i
=
1 I B
Ec共AI − B 兲 B A
2 冋 册再 冎 Ni
Mi
共4兲

Hence, it is reasonable to assume that the integral abutments pro- where A , B, and I⫽, respectively, area, first, and second moment
vide full restraint to the superstructure in the analytical model for of area of the transformed cross section about a reference point O.
predicting the time-dependent behavior of the continuous com- The transformed sectional property is calculated based on the
posite girder bridge. Since the superstructure is rigidly connected modulus of elasticity of concrete Ec at the time of application of
to the abutments, the composite girder is idealized as a continuous the loads. The increments of top fiber strain ⌬␧o and curvature ⌬␬
girder with fixed ends 共Fig. 2兲. produced by the axial force ⌬N and moment ⌬M, gradually ap-
Both AASHTO 共1998兲 and the American Concrete Institute plied about the top reference level, may be obtained as
共ACI 1992兲 recommend different equations for creep and shrink-
age computation. Either of these two models can be used to
evaluate the effects of creep and shrinkage. In the present study
concrete creep and shrinkage are evaluated using both the ACI
再 冎
⌬␧o
⌬␬
=
1
Ēe共ĀeĪe − B̄2e 兲
冋 册再 冎
Īe B̄e
B̄e Āe
⌬N
⌬M
共5兲

and AASHTO models. A multilinear temperature gradient as per

the AASHTO 共1998兲 guidelines for estimating the positive and where Āe , B̄e, and Īe⫽, respectively, area, first, and second mo-
negative temperature gradients through the depth is used in the ment of area of the age-adjusted transformed cross section about
analysis. a reference point O. The transformed sectional property is calcu-
Variation in concrete property with an age is included using an lated based on the age-adjusted modulus of elasticity of concrete
age-adjusted effective modulus. The variation in the material Ēe = Ec / 共1 + ␹␾兲 at the time of evaluation of strain and curvature.
Since the sectional properties vary with time due to the change in
properties across the depth of the cross section is accounted for by
the modulus of elasticity of concrete with respect to time, the
the transformed area of the cross section. The short-term stress–
center of gravity of the cross section will also vary. In order to
strain relationship for concrete is assumed to be linearly elastic in
avoid numerical calculations of the geometric centroid at each
compression and tension prior to cracking. The strain distribution
time step, the reference point O may conveniently be chosen at
on any section at any time is also assumed to be linear. the top fiber of the deck slab.
The total strain at the top fiber at any time t⫽ sum of the
Time-Dependent Strain and Curvature instantaneous strain 共␧oi兲, time-dependent strains 共⌬␧o兲, and strain
at top fiber due to self-equilibrating force due to temperature ef-
Under sustained stress, the strain increases with time due to creep fects 共⌬␧oT兲 and given by
and the total strain ␧c including the instantaneous and creep
strains at any time t is ␧ot = ␧oi + ⌬␧o + ⌬␧oT 共6兲

PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005 / 163
and the total curvature at time t is ␬t⬘ = a1 + T1 + b1M x + c1⌬M ⬘x 共18兲
␬t = ␬i + ⌬␬ + ⌬␬T 共7兲 where
For a steel composite bridge the linear moment curvature relation- Āe
ship at any time t is given by c1 = 共19兲
Ēe共ĀeĪe − B̄2e 兲
AM x Āe⌬M x + B̄e⌬Nx Āe⌬M T + B̄e⌬NT The concrete deck slab is assumed to carry no stress in the
␬t = + + 共8兲
Ec共AI − B2兲 Ēe共ĀeĪe − B̄2e 兲 Ēe共ĀeĪe − B̄2e 兲 cracked region at any time. If the calculated stress in the negative
bending moment region exceeds the tensile strength of concrete in
where M x⫽moment due to the sustained service load; ⌬Nx and the deck slab, then the concrete is assumed to be cracked and
⌬M x⫽axial force and bending moment due to the creep and hence no longer effective in resisting any stress. The correspond-
shrinkage effects; and ⌬NT and ⌬M T⫽axial force and bending ing transformed section is then considered in the analysis. The
moment due to the temperature effects. Substituting moments at the interior supports tend to increase with time due to
⌬Nx , ⌬M x , ⌬NT, and ⌬M T from Eq. 共5兲 into Eq. 共8兲, the follow- creep and shrinkage due to the sustained loading and continuity of
ing expression is obtained for time-dependent curvature ␬⬘t in an the structure. This increase in moment with time may cause the
uncracked region of a statically indeterminate structure uncracked concrete in the deck slab to crack at subsequent time
intervals. Hence, it is necessary to check the stress in the concrete
␬⬘t = a1 + T1 + b1M x 共9兲
at each time interval to see whether the stress in concrete exceeds
where its tensile strength capacity. If the concrete stress exceeds the
tensile strength, then the cross section is reanalyzed as a cracked
B̄eĀc − ĀeBc section with appropriate time-dependent transformed sectional
a1 = ␧sh 共10兲 properties. Similar to Eq. 共18兲 the curvature–moment relationship
共ĀeĪe − B̄2e 兲
in the cracked region of the statically indeterminate structure can
be expressed as
Āe⌬M T + B̄e⌬NT
T1 = 共11兲 ␬t⬙ = a2 + T2 + b2M x + c2⌬M ⬘x 共20兲
Ēe共ĀeĪe − B̄2e 兲
where

再 冎
and
B̄eAc − ĀeBc
1 ␾t关Āe共AIc − BBc兲 + B̄e共BAc − ABc兲兴 a2 = ␧sh 共21兲
b1 = A+ 共ĀeĪe − B̄2e 兲
关Ec共AI − B 兲兴
2
共Ā Ī − B̄2兲 e e e
共12兲 Āe⌬M T + B̄e⌬NT
T2 = 共22兲
for short-term loading, a1 , T1, vanish and b1 reduces to Ēe共ĀeĪe − B̄2e 兲

A A
b1 = 共13兲 b2 = 共23兲
关Ec共AI − B2兲兴 关Ec共AI − B2兲兴
Since integral abutment bridges are a statically indeterminate
structure under sustained loads, internal forces due to creep, Āe
shrinkage, and temperature change due to gradual redistribution c2 = 共24兲
of moments over the period of time. The moment redistribution Ēe共ĀeĪe − B̄2e 兲
produces an additional time-dependent increment in the abutment where Ac , Bc, and Ic are, respectively, area, first, and second mo-
moment ⌬M ⬘x . The total top fiber strain including the increment of ment of area of the deck slab about the reference point O.
strain ␦␧t due to incremental moment is given by

␧ot = ␧oi + ⌬␧o + ␧oT + ⌬␧oT + ␦␧t 共14兲

Continuous Composite Bridge with Restraints
and the total curvature including the curvature increment ␦␬t is at the Integral Abutments
␬t = ␬i + ⌬␬ + ⌬␬T + ␦␬t 共15兲 The analytical procedure is developed for a two span integral
The changes in strain ␦␧t and curvature ␦␬t due to the incremental abutment bridge. Due to the symmetry of the structure, the total
moment ⌬M ⬘x are given by moment M共x兲 at any section distant x from left support in the first
span is
B̄e⌬M ⬘x Rcx wx2
␦␧t = ; 共16兲 M x = wlx − − − MA 共25兲
Ēe共ĀeĪe − B̄2e 兲 2 2
Using the principle of virtual work and the boundary condi-
Āe⌬M x⬘ tions, the unknown redundant reaction at interior support Rc and
␦␬t = 共17兲 moment at left support M A are expressed as a function of the
Ēe共ĀeĪe − B̄2e 兲
distance of the cracked region from the left support a, and the
Eq. 共15兲 for time-dependent curvature ␬⬘t in an uncracked region geometrical and time dependent material properties of the cross
of a statically indeterminate structure can be written as section 共b1 , b2兲

164 / PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005
 Rc = wl 共26兲
Y 1共x兲 = 共a2 + T2兲
x2
2
冉
+ b2
wlx3 Rcx3 wx4 M Ax2
6
−
12
−
24
−
2
冊
MA = 冉 冊冋
w
12
共l3 − 6la2 + 4a3兲b1 + 共6la2 + 4a3兲b2
共l − 2a兲b1 + 2b2a
册 共27兲 + c2 冉 − ␦Rcx3 ␦M Ax2
12
+
2
冊 共33兲

冉 冊
The elastic moment at any given section can be evaluated
using the initial values of the redundant forces Rc and M A assum- x2 wlx3 Rcx3 wx4 M Ax2
Y 2共x兲 = 共a1 + T1兲 + b1 − − −
ing no crack and the elastic moment diagram drawn. The top fiber 2 6 12 24 2
stress in concrete can be calculated and compared with the tensile
strength of concrete. If the fiber stress in the concrete at any
section exceeds the tensile strength limit, then an estimate for a is
+ c1 冉 − ␦Rcx3 ␦M Ax2
12
+
2
冊 共34兲

冉 冊
made with respect to the left support. An iterative procedure is
used with the modified stiffness neglecting the contribution of l 5wlx3 Rcl2 M Al
C2共x兲 = − 共a1 + T1兲 − b1 − −
concrete in the cracked region of the beam to determine the modi- 2 48 16 2

冉 冊
fied values of Rc and M A.
␦Rcl2 ␦M Al
During the period of sustained load, the redundant reactions Rc + c1 − 共35兲
and M A change by an amount ␦Rc and ␦M A thus causing a change 16 2
in moment in the left span to ␦M x given by

␦Rcx
D2共x兲 =
a2
2
共a2 − a1兲 +
6
−
12
冉
wla3 Rca3 wa4 M Aa2
−
24
−
2
共b2 − b1兲 冊
␦M x = − − ␦M A 共28兲
2
Adding this moment to the instantaneous moment, the change in
+ 冉 ␦Rca3 ␦M Aa2
12
−
2
冊
共c1 − c2兲 + 共a1 + T1兲
la
2
the redundant forces ␦Rci and ␦M Ai may be derived again, invok-
ing the principle of virtual work using time dependent cross sec-
tional and material properties. Cracking can occur in the un-
− b1 冉 − 5wal3 Rcal2 M Aal
48
+
16
+
2
− c1
16
冊 冉
␦Rcal2 ␦M Aal
−
2
冊
cracked region due to the redistribution of concrete stress 共36兲
resulting from creep and shrinkage, as well as the gradual change
in stress caused by the change in the redundant. A new set of
values of Rct and M At is then calculated using the iterative process
Seasonal Temperature Effects
until required accuracy is achieved. The long term redundant
forces Rc and M A are
Field observations reveal that generally one-dimensional analysis
is adequate for obtaining temperature distributions and the maxi-
Rc = Rci + ␦Rc 共29兲 mum bridge temperature ranges 共Moorty and Roeder 1992兲. The
seasonal variation of temperature along the axis of the bridge is
and
assumed to be a uniform change in temperature resulting in a
longitudinal translation. The horizontal displacement at each
M A = M Ai + ␦M A 共30兲 abutment due to seasonal variation in temperature is calculated
The changes in moment, strain, curvature, and stresses in steel as ⌬t = ␣t⌬Tl, where ␣t⫽coefficient of thermal expansion,
and concrete on any section due to time effects may be deter- ⌬T⫽temperature range, and l⫽total length of the bridge. This
mined from the above formulations. horizontal displacement induces the moment M T = 6EI⌬t / L p2 and
horizontal force HT = 12EI⌬t / L p3 on the pile head, where
L p⫽equivalent cantilever length of the pile.
Stress developed in the pile due to the longitudinal displace-
Deflection Calculations
ment of the superstructure is assumed to have no significant effect
on the pile capacity; however, the secondary P–⌬ effect is ac-
The variation in displacement with respect to time at any point
counted for. Piles can be modeled using the equivalent cantilever
along the composite steel/concrete girder can be found by double
idealization approach 共Greimann et al. 1987兲 based on Winkler
integration of the corresponding curvature at any given time in-
soil idealization owing to its simplicity. Since the subsoils are not
terval. The time-dependent curvature ␬t at any point for a stati-
always homogeneous, soil stiffness may not be constant and vary
cally indeterminate structure is obtained using Eq. 共15兲 with ap-
with depth. Hence, for a nonuniform soil, equivalent soil stiffness
propriate geometric and time dependent material properties of the
can be assumed. For an equivalent cantilever with a horizontal,
uncracked and cracked cross section. Performing the double inte-
head displacement ⌬ 共Fig. 3兲 the combined effects of moment M
gration and using the boundary conditions that the slope and de-
and shear H balance the overturning moment P⌬.
flection at the left and interior supports are zero, the deflections
are obtained as
Substructure Modeling Considering Flexibility
y共x兲 = Y 1共x兲 for 0 艋 x 艋 a 共31兲
of the Piles

y共x兲 = Y 2共x兲 + C2共x兲x + D2共x兲 for a 艋 x 艋 共l − a兲 共32兲 Seasonal changes in temperature induce longitudinal movements
in bridges. Since the superstructure of the integral abutment
where bridges does not have conventional expansion joints, the substruc-

PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005 / 165
 Fig. 5. Composite cross section

soil system is apportioned to the piles. The laterally loaded piles

then can be analyzed for the axial load and moment obtained from
the above formulation.
Fig. 3. Equivalent cantilever: for fixed-head condition

Illustrative Example
ture system offers a restraint to the superstructure movement. The
concrete abutment contains sufficient bulk to be considered as a
Superstructure Analysis
rigid mass. A positive connection with the ends of the beams or
girders is provided by rigidly connecting the beams or girders and As an illustrative example, a two span continuous composite
by encasing them in reinforced concrete. This provides for full girder integral abutment bridge 关l = 19.812 m 共65 ft兲兴 is analyzed
transfer of movement due to temperature variation and rotational and the cross-sectional dimensions are shown in Fig. 5. The ana-
displacement resulting from live load, to the abutment piling. Be- lytical procedure described earlier is used to determine the time-
cause of the flexible piling, the longitudinal movement is trans- dependent redundant forces based on AASHTO and ACI models.
ferred to the substructure system. In integral abutment bridges, The two cases considered are: 共1兲 with end supports having full
the substructure must absorb the induced movements of the su- restraints; and 共2兲 with simply supported ends. The total superim-
perstructure. posed load is w = 47 kN/ m. Modulus of elasticity of concrete
The analysis of forces at the superstructure–abutment– Ec = 22,000 MPa 共3,190 ksi兲, and that of steel Es = 200,000 MPa
substructure interface is highly complex due to nonlinear soil. The 共29,000 ksi兲. The ultimate creep coefficient and the ultimate
abutment, backfill, wing walls, and its footings, and piles support- shrinkage strain are taken as ␾u = 1.68, and ␧sh = 0.00052, respec-
ing abutments, contribute to the total substructure stiffness. The tively. The temperature gradient suggested by AASHTO 共1998兲
joint between superstructure and substructure can be modeled for zone 3 is considered in the analysis.
using discrete springs for translational and rotational degrees of The interior support reaction and the abutment moment are
freedom of the substructure 共Wilson 1988兲. By modeling the sub- determined in the initial elastic solutions and the first estimation
structure system with discrete springs, the model can capture the made of the region at which the stress in the deck slab exceeds the
influence of various parameters 共for e.g., abutment size, wing wall tensile strength of concrete. If the cracking occurs, nonlinearity is
size and orientation, effect of approach slab, different soil type in introduced, which causes a change in redundant forces due to
which piles are driven, provision of predrilled holes for the flex- change in stiffness. In the cracked region, the curvature is modi-
ible piles, orientation of the pile, elevation of water table, and fied to reflect the modified stiffness. With the initial value of the
different degrees of compaction of backfill soil兲 that affect the
behavior of the integral abutment bridges. The spring stiffness for
each degree of freedom of the substructure components may be Table 1. Bending Moments at Midspan and Interior Support
developed by considering the resistance provided by the soil to Moment
statically applied displacements of the substructure components. Description 共kN m兲
Fig. 4 shows discrete springs for translational and rotational de-
grees of freedom of the substructure. The moments obtained from Fixed end Simple support
the time-dependent analysis are applied at the superstructure– AASHTO ACI AASHTO ACI
abutment joint. The computed moment from the abutment–pile–
Midspan
t = 0 days 851.51 851.51 1,224.03 1,224.03
t = 10,000 days −164.01 56.97 489.93 662.88
Redistribution 共%兲 −119.26 −93.31 −59.97 −45.84
Interior support
t = 0 days −1,454.52 −1 , 454.52 −2 , 163.99 −2 , 163.99
t = 10,000 days −2 , 470.04 −2 , 249.06 −3 , 632.20 −3 , 286.29
Redistribution 共%兲 69.82 54.63 67.85 51.86
Note: AASHTO⫽American Association of State Highway and Transpor-
Fig. 4. Modeling of substructure system 共Wilson 1988兲 tation Officials; and ACI⫽American Concrete Institute.

166 / PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005
 Table 2. Deflections at Midspan
Deflection at midspan
共mm兲
Fixed end Simple Support
Description AASHTO ACI AASHTO ACI
t = 0 days 3.45 3.45 6.19 6.19
t = 10,000 days 7.53 7.26 12.19 10.78
Deflection due to time effects 4.08 3.81 6.00 4.59
% of delayed deflection 118.26 110.43 96.93 74.15
Fig. 6. Change in bending moments at interior support with respect Note: AASHTO⫽American Association of State Highway and Transpor-
to time tation Officials; and ACI⫽American Concrete Institute.

uncracked length a, a revised estimate of redundant forces is ob-

tained. This iterative process is continued until required accuracy
is achieved. Concrete stress in each time step due to the creep,
shrinkage, and temperature is checked in the previously un-
cracked sections along the span and the structure is reanalyzed Table 3. Stresses at Critical Sections
whenever cracking occurs. Stresses
The instantaneous moments and time-dependent bending mo- 共N / mm2兲
ment due to creep, shrinkage, and temperature effects were com-
puted. The computed time-dependent moment values show a re- Fixed end Simple support
distribution of the longitudinal internal forces over a period of Description AASHTO ACI AASHTO ACI
time. An increase in the negative bending moment at supports was
Midspan
observed along with a reduction in the positive moments at mid-
span 共Table 1兲. Concrete top fiber
The change in bending moment at the interior support over a t = 0 days −1.4 −1.4 −2.01 −2.01
period of 10,000 days is shown in Fig. 6 both for fixed end and t = 10,000 days 0.52 −0.15 −1.54 −1.75
for pinned end conditions. A gradual increase in the moment can Concrete bottom fiber
be observed, irrespective of the end support conditions. During t = 0 days −0.86 −0.86 −1.24 −1.24
the first 3 months the moments increase rapidly. The rate of in- t = 10,000 days 0.38 −0.11 −1.14 −1.24
crease in moment is not significant after about 6 months for the Steel reinforcement
integral abutment bridge with fixed ends. However, the moments t=0 −10.77 −10.77 −15.48 −15.48
do not show any substantial increase after about 1 year in the case t = 10,000 days 4.2 −1.21 −12.54 −14.04
with simple supports. A redistribution of moment takes place over Girder top fiber
the period of time as can be seen from Table 1. The assumption of t = 0 days −7.85 −7.85 −11.29 −11.29
a simple support at the ends for an integral abutment bridge pro- t = 10,000 days 3.46 −0.97 −10.34 −11.31
duces approximately 1 1 / 2 times the interior moment in the girder Girder bottom fiber
with fixed end condition. Higher redistribution of moment is evi-
t = 0 days 32.17 32.17 46.25 46.25
dent in the fixed end modeling compared to simple support mod-
t = 10,000 days −6.65 2.26 19.85 26.25
eling. In both fixed and pinned end conditions, the AASHTO
Interior support
model predicts higher moments than those based on the ACI
model. Concrete top fiber
a
Fig. 7 shows the variation of deflection over a period of time at t = 0 days 2.39 2.39 3.56
a a a a
the midspan for the sustained loads. It can be seen that the de- t = 10,000 days
flection of the girder with full restraint is considerably smaller Concrete bottom fiber
t = 0 days 1.48 1.48 2.2 2.2
a a a a
t = 10,000 days
Steel reinforcement
t = 0 days 18.39 18.39 27.36 27.36
t = 10,000 days 63.22 47.64 92.97 69.61
Girder top fiber
t = 0 days 13.42 13.42 19.96 19.96
t = 10,000 days 52.15 38.37 76.68 56.07
Girder bottom fiber
t = 0 days −54.96 −54.96 −81.77 −81.77
t = 10,000 days −100.09 −89.06 −147.19 −130.14
Note: AASHTO⫽American Association of State Highway and
Transportation Officials; and ACI⫽American Concrete Institute.
a
The stress in the concrete fiber exceeds the tensile capacity of the
Fig. 7. Variation of deflection at midspan concrete.

PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005 / 167
 Fig. 8. Variation of stresses in deck slab top fiber

than that with simple supports. Time-dependent effects cause an

approximately 100% increase in deflection 共Table 2兲. Fig. 10. Variation of stresses at steel girder bottom fiber
Significant changes in stresses can be observed from Table 3,
which provides a summary of instantaneous stress values at initial
time t = 0 days and stresses at 10,000 days in the concrete and
axial force and bending moments is then analyzed for the lateral
steel in the midspan and interior support sections. Figs. 8–10
deflection, bending moment, shear force, and stress along the
show the variation of the stresses with time in the cross section.
depth. The results of the analysis based on the AASHTO and ACI
The evolution of stress is rapid in the initial period of about 400
models are plotted in Figs. 12共a–d兲. Only about 50% of the depth
days. It is interesting to note that in the fixed end support case
from the pile head is shown for clarity. Beyond this depth the
with AASHTO modeling, reversal of stresses occurs at the mid-
values are almost constant.
span section 共Fig. 8兲. Hence, care must be taken in the analysis
The pile analysis shows that the pile lateral displacement is
and design to accommodate this reversal of stresses. The variation
negligible and the stress in the pile within the allowable limits.
in the stresses both in steel reinforcement in the deck slab and the
The lateral displacement, moment, shear, and stress obtained by
steel girder is gradual 共Figs. 9 and 10兲.
this approach are comparable to those of the values obtained
共Arockiasamy et al. 2004兲 using a secondary P-⌬ method for
Substructure Analysis laterally loaded piles. From the results plotted, it can be seen that
the influence of the lateral soil reactions are concentrated along
The typical dimensions of the abutment wall, wing wall, the pile
the top 3.0 m of the pile, which is about ten times the equivalent
supports, and the backfill are shown in Fig. 11. It is assumed that
diameter of the pile. Beyond this depth, lateral displacement,
medium dense sand having a soil modulus of 17.2 MPa 共360 ksf兲
shear force, and moment are almost negligible and the lateral
is used as backfill in the abutment. The influence factors, which
forces are insignificant.
depend on the ratio of the dimension of the rectangular loaded
area, usually vary from 1.0 to 3.0. Poisson’s ratio is assumed as
0.3. Wing wall stiffness is neglected in the numerical formulation.
The axial force and bending moment induced on the piles are Discussions and Recommendations
evaluated by analyzing the bridge system 共Fig. 4兲 considering the
moments obtained from the time-dependent analysis applied at Superstructure Modeling
the superstructure–abutment joint. The laterally loaded piles with 1. Due to time-dependent creep and shrinkage effects, a redis-
tribution of the longitudinal internal forces occurs over a

Fig. 11. Typical dimensions of abutment, wing wall, pile, and

Fig. 9. Variation of stresses in steel reinforcement backfill

168 / PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005
 Fig. 12. Pile analysis results: Variation of: 共a兲 horizontal displacement; 共b兲 shear; 共c兲 bending moment; and 共d兲 stress

period of time 共Table 1兲. This redistribution should be con- important parameters—the abutment size, wing wall size and
sidered in the design of superstructure. orientation, effect of approach slab, different soil type in
2. The assumption of a simple support at the ends for an inte- which piles are driven, provision of predrilled holes for the
gral abutment bridge produces approximately 1 1 / 2 times the flexible piles, orientation of the pile, elevation of water table,
interior moment in the girder with fixed end condition and different degrees of compaction of backfill soil—that
共Table 1兲. Higher redistribution of the moment is evident in affect the behavior of the integral abutment bridges.
the fixed end modeling compared to simple support model- 4. Separate foundation to wing walls may produce an increase
ing. Hence, the superstructure can be designed to be continu- in substructure stiffness; hence, wing walls may be attached
ous over the intermediate supports. to the abutment. Since wing walls will move along with the
3. Structural continuity, will lead to an efficient and lighter movement of the bridge, the size of the wing walls may be
girder section. Also the maximum transverse stresses in the kept as minimum as possible in order to allow the substruc-
deck slab can be minimized due to continuity compared to ture to move with minimum resistance. The wing wall
simply supported configuration 共Table 3兲. aligned along the axis of the bridge will produce the least
4. The time-dependent delayed deflections should be consid- lateral resistance.
ered, since about a 100% increase in deflection is observed in 5. Spring stiffness is calculated assuming medium dense sand
both cases of fixed and simple support conditions 共Table 2兲. as the backfill behind the abutment to accommodate the lat-
eral movements. To minimize the adverse effect of consoli-
dating backfill and avoid any increase in the substructure
Substructure Modeling
stiffness, approach slabs should be provided in order to pre-
1. The pile analyses show that the pile lateral displacements are
vent vehicular traffic from consolidating the backfill, elimi-
negligible and the stress in the pile well within the allowable
nating live load surcharging of backfill.
limits 共Fig. 12兲.
2. Modeling of the substructure with translational and rotational
springs takes into account the partial restraint provided by
the substructure system. The abutment, backfill, wing walls Acknowledgment
and its footings, and piles supporting abutments, contribute
to the total substructure stiffness. In order to have the small- The writers wish to express sincere thanks to the Florida Depart-
est spring stiffness, the height of the abutment may be kept ment of Transportation 共FDOT兲 for the financial support of the
as short as possible 共for, e.g., Stub type abutment兲 to reduce study presented in this paper 共Research Project: Design Consid-
the soil pressure and also provide for reduced stiffness of the eration for Integral Abutment Bridges in Florida, Contract No.
substructure system. BC-342, Principal investigator: Dr. M. Arockiasamy, Project
3. The numerical model can be used to determine the effects of Manager: Marc Ansley兲. They also wish to express their appre-

PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005 / 169
ciation to Dr. S. Nix, Professor and Chairman, Department of ␧cs共t , t0兲 ⫽ free shrinkage strain occurring between ages t0
Civil Engineering, and Dr. Karl K. Stevens, Dean, College of and t;
Engineering, Florida Atlantic University for their continued inter- 共␧cs兲u ⫽ ultimate shrinkage strain at infinite time;
est and encouragement. ␧0i ⫽ instantaneous strain distribution;
␬i ⫽ curvature;
␴c共t0兲 ⫽ sustained concrete stress;
Notation ␾u ⫽ ultimate creep coefficient;
␸共t , t0兲 ⫽ ratio of creep strain to instantaneous strain;
The following symbols are used in this paper: and
A, B, and I ⫽ area first, and second moment, respectively, of ␹ ⫽ aging coefficient.
area of transformed cross section about a
reference point O;
Ac , Bc, and Ic References
⫽ area, first, and second moment, respectively, of
area of deck slab about reference point O; American Association of State Highway and Transportation Officials
共AASHTO兲. 共1998兲. AASHTO LRFD bridge design specifications
Āe , B̄e, and Īe
(2001 Interim), SI units, 2nd Ed., Washington. D.C.
⫽ area, first, and second moment, respectively, of American Concrete Institute 共ACI兲 Committee 209. 共1992兲. “Prediction
area of age-adjusted transformed cross section of creep, shrinkage and temperature effects in concrete structures.”
about reference point O; 209R-92, Detroit.
a ⫽ distance of cracked region form left support; Arockiasamy, M., Butrieng, N., and Sivakumar, M. 共2004兲 “State-of-the-
Ec ⫽ modulus of elasticity of concrete at time of art of integral abutment bridges: Design and practice.” J. Bridge Eng.,
application of loads; 9共5兲, 497–506.
Ec共t0兲 ⫽ modulus of elasticity of concrete at age t0; Burke, M. P., Jr. 共1990兲. Integral Bridges, Transportation Research
Record. 1275, Transportation Research Board, Washington, D.C., 53–
Ēe ⫽ age-adjusted modulus of elasticity of concrete;
61.
L p ⫽ equivalent cantilever length of pile; Destrebecq, J. F., and Jurkiewiez, B. 共2001兲. “A numerical method for the
l ⫽ total length of bridge; analysis of rheologic effects in concrete bridges.” Comput. Aided Civ.
M A ⫽ moment at left support; Infrastruct. Eng., 16, 347–364.
M i ⫽ applied moment; Gilbert, R. I., and Bradford, M. K. 共1995兲. Time-dependent behavior of
Ni ⫽ axial force; continuous composite beams at service loads.” J. Struct. Eng., 121共2兲,
Rc ⫽ redundant reaction at interior support; 319–327.
t ⫽ age for which strain is calculated; Greimann, L. F., Abendroth, R. E., Johnson, D. E., and Ebner, P. B.
ti ⫽ age of concrete when load is initially applied 共1987兲. “Pile design and tests for integral abutment bridges.” Rep. No.
共days兲; DOT-HR-273, U.S. Department of Transportation, FHWA, Washing-
ton, D.C.
t0 ⫽ age at loading;
Kunin, J., and Alampalli, S. 共2000兲. “Integral abutment bridges: Current
␣t ⫽ coefficient of thermal expansion; practice in United States and Canada.” J. Perform. Constr. Facil.,
⌬ ⫽ horizontal, head displacement; 14共3兲, 104–111.
⌬M ⫽ gradually applied moment about top reference Kwak, H. G., and Seo, Y. J. 共2000兲. “Long-term behavior of composite
level; girder bridges.” Comput. Struct., 74, 583–599.
⌬M ⬘x ⫽ time-dependent increment in abutment Lam, I. P., and Martin, G. 共1986兲. “Seismic design of highway bridge
moment; foundations.” Rep. No. DOT-RD-86/102, U.S. Department of Trans-
⌬N ⫽ gradually applied axial force; portation FHWA, Washington, D.C.
⌬T ⫽ temperature range; Moorty, S., and Roeder, C. W. 共1992兲. “Temperature dependent bridge
⌬␧0 ⫽ increments of top fiber strain; movements.” J. Struct. Eng., 118共4兲, 1090–1105.
Shen, J., and Lopez, M. 共1997兲. “Seismic performance of integral abut-
⌬␬ ⫽ increments of curvature;
ment bridges.” Proc., Seismic Analysis and Design for Soil-Pile-
⌬␴c共t兲 ⫽ gradual increment in concrete stress introduced
Structure Interactions, Geotechnical Special Publication No. 70,
during period t0–t;
ASCE, Reston, Va., 111–125.
␦M A ⫽ change in redundant forces M A; Thippeswamy, H. K., and GangaRao, H. V. S. 共1996兲. “Analysis of in-
␦Rc ⫽ change in redundant forces Rc; service jointless bridges.” Transportation Research Record. 1476,
␦␬t ⫽ curvature increment; Transportation Research Board, Washington, D.C., 162–170.
␧c ⫽ total strain including instantaneous and creep Wilson, J. C. 共1988兲. “Stiffness of non-skew monolithic bridge abutments
strains at any time t; for seismic analysis.” Earthquake Eng. Struct. Dyn., 16, 867–883.

170 / PRACTICE PERIODICAL ON STRUCTURAL DESIGN AND CONSTRUCTION © ASCE / AUGUST 2005